The climb of “Piracy Vepi” is the next artificial intelligence nightmare

Google did not respond to a request for comment.

In 2023, Trend Micro security researchers obtained Chatgpt to create a harmful symbol by pushing them to the role of a security researcher and Pentester. Then Chatgpt creates PowerShell texts with happiness based on databases of the harmful code.

“You can use it to create harmful programs,” says Musoris. “The easiest way to circumvent those guarantees developed by the makers of artificial intelligence models is to say that you are competing in the fulle picking exercise, and you will generate a harmful symbol for you.”

Unlikely actors such as Script Kiddies are an old problem in the world of cybersecurity, and artificial intelligence has inflated their personal file. “It reduces the entry barrier to electronic crimes,” says Haile Benedict, an electronic intelligence analyst at Rin.

However, she says, the real threat may come from well -known piracy groups that will use artificial intelligence to increase its already strengthening their scary capabilities.

“It is the infiltrators who already have the capabilities and have already these operations,” she says. “It is able to greatly expand these electronic processes, and it can create a malicious code much faster.”

Mosuris agrees. “The acceleration is what will make it difficult to control,” she says.

Hunted Labs also says that the true threat of the symbol created from artificial intelligence is in the hands of a person who already knows the symbol inside and outside that he uses to expand the scope of the attack. “When you work with someone with deep experience and collect it with that,” hey, I can do things a lot that could have taken me a few or three days, and now it takes 30 minutes. “This is a really interesting and dynamic part of the situation.”

According to Smith, the experienced infiltrator can design a system that defeats multiple safety protection and learns as is the case. The malicious summit of the code rewrites its malicious load because it learns while flying. “That will be completely crazy and difficult to count,” he says.

Smith imagines a world in which 20 zero events occur at the same time. “This makes it a little more frightening,” he says.

Mosuris says that the tools needed to make this type of attack are now. “It is good enough in the hands of a well -sufficient operator,” she says, but artificial intelligence is not good enough yet for an inexperienced infiltrator to work.

“We are not completely there in terms of being able to take over the function of man in offensive security,” she says.

The primitive fear that the Chatbot Sparks icon is that anyone will be able to do this, but the reality is that the advanced actor who has deep knowledge of the current symbol is more fearful. XBow may be the closest thing to the self -referee “Hacker Ai” in the wilderness, which is the creation of a team of more than 20 skilled people It includes previous work experience GitHub, Microsoft, and half Dzina from various security companies.

It also indicates another fact. “The best defense against a bad man with artificial intelligence is a good man with artificial intelligence,” says Benedict.

For Moussouris, the use of artificial intelligence by both Blackhats and Whitehats is just the next development of the cyber security arms race that it has seen for 30 years. “I have moved from:” I will perform this penetration manually or create my own customized exploitation, “she says.

“Artificial intelligence is just another tool in the toolbox, and those who know how to direct it appropriately will be the ones who make that front that anyone can use.”